diff --git a/lib/modules/String.pmod/module.pmod b/lib/modules/String.pmod/module.pmod
index 2696d935af9630f5e6992455f493ee9ebeb3986f..8fef01347ed1aaa198a39717fac10abe9b1e0f3c 100644
--- a/lib/modules/String.pmod/module.pmod
+++ b/lib/modules/String.pmod/module.pmod
@@ -6,6 +6,7 @@ constant Buffer = __builtin.Buffer;
constant count=__builtin.string_count;
constant width=__builtin.string_width;
constant trim_whites = __builtin.string_trim_whites;
+constant normalize_space = __builtin.string_normalize_space;
constant trim_all_whites = __builtin.string_trim_all_whites;
constant Iterator = __builtin.string_iterator;
constant SplitIterator = __builtin.string_split_iterator;
diff --git a/lib/modules/String.pmod/testsuite.in b/lib/modules/String.pmod/testsuite.in
index 9e939c0c7dba205739d2d8a840f26e197fae778d..e5eb04661f359b1567acc8f0996e464beaf24136 100644
--- a/lib/modules/String.pmod/testsuite.in
+++ b/lib/modules/String.pmod/testsuite.in
@@ -1,5 +1,5 @@
START_MARKER
-dnl $Id: testsuite.in,v 1.26 2008/06/29 18:03:03 nilsson Exp $
+dnl $Id: testsuite.in,v 1.27 2010/01/02 13:17:10 srb Exp $
test_eq([[ String.Buffer()->add("xxx") ]], 3)
test_any([[
@@ -116,6 +116,25 @@ test_eq(String.trim_all_whites ("\200000"), "\200000")
test_eq(String.trim_all_whites (" \t\n\r "), "")
test_eq(String.trim_all_whites (""), "")
+test_eq(String.normalize_space ("\v\f \t\n\r\0\v\f \t\n\r"), "\0")
+test_eq(String.normalize_space ("\v\f \t\n\r\400\v\f \t\n\r"), "\400")
+test_eq(String.normalize_space ("\v\f \t\n\r\200000\v\f \t\n\r"), "\200000")
+test_eq(String.normalize_space ("\v\f \t\n\ra\400\v\f \t\n\ra\400\t \v\f"),
+ "a\400 a\400")
+test_eq(String.normalize_space ("\0"), "\0")
+test_eq(String.normalize_space ("\400"), "\400")
+test_eq(String.normalize_space ("\200000"), "\200000")
+test_eq(String.normalize_space (" \t\n\r\v\f "), "")
+test_eq(String.normalize_space (""), "")
+test_eq(String.normalize_space (" a bb ccc ddd \n eee f g\n"),
+ "a bb ccc ddd eee f g")
+test_eq(String.normalize_space (" a bb ccc ddd \n eee f g\n"," \t"),
+ "a bb ccc ddd\neee f g\n")
+test_eq(String.normalize_space (" a bb ccc ddd \n eee f g\n","\t "),
+ "a\tbb\tccc\tddd\neee\tf\tg\n")
+test_eq(String.normalize_space (" a bb ccc ddd \n eee f g\n",""),
+ " a bb ccc ddd \n eee f g\n")
+
dnl MISSING TEST: String.Iterator
test_any([[
diff --git a/src/builtin.cmod b/src/builtin.cmod
index 194ef65ce9c59e10e436adb9a9e090528d9769d3..c2795e06a36b1a9412dc7212c8045bb6eec8c7bb 100644
--- a/src/builtin.cmod
+++ b/src/builtin.cmod
@@ -2,7 +2,7 @@
|| This file is part of Pike. For copyright information see COPYRIGHT.
|| Pike is distributed under GPL, LGPL and MPL. See the file COPYING
|| for more information.
-|| $Id: builtin.cmod,v 1.241 2009/11/19 23:45:22 mast Exp $
+|| $Id: builtin.cmod,v 1.242 2010/01/02 13:17:10 srb Exp $
*/
#include "global.h"
@@ -750,6 +750,106 @@ PIKEFUN string string_trim_whites (string s)
RETURN string_slice (s, start, end + 1 - start);
}
+/*! @decl string normalize_space (string s, string|void whitespace)
+ *! @belongs String
+ *!
+ *! Returns @[s] with white space normalised.
+ *! White space is normalised by stripping leading and trailing white space
+ *! and replacing sequences of white space characters with a single space.
+ *! @[whitespace] is defined to be " \t\r\n\v\f" if omitted, the first
+ *! character denotes the replacement character for replacing sequences.
+ *!
+ *! Note that trailing and leading whitespace around \r and \n characters
+ *! is stripped as well.
+ */
+PMOD_EXPORT
+PIKEFUN string string_normalize_space (string s, string|void whitespace)
+ errname String.normalize_space;
+ optflags OPT_TRY_OPTIMIZE;
+{ size_t len = s->len;
+ void *src = s->str;
+ unsigned shift = s->size_shift;
+ const char *ws;
+ struct string_builder sb;
+ unsigned foundspace = 0;
+
+ if(whitespace)
+ if(whitespace->size_shift>8)
+ Pike_error("Cannot use wide strings for whitespace\n");
+ else if(!whitespace->len)
+ REF_RETURN s;
+ else
+ ws = whitespace->str;
+ else
+ ws = 0;
+
+ init_string_builder_alloc (&sb, len, shift);
+ sb.known_shift = shift;
+#define DO_IT_SPACECHECK(c) \
+ ((c)==' '||(c)=='\t'||(c)=='\r'||(c)=='\n'||(c)=='\v'||(c)=='\f')
+ switch (shift) {
+#define DO_IT(TYPE) \
+ { TYPE *start = src, *end = start+len, *dst = (void*)sb.s->str; \
+ for (; start < end; start++) { \
+ unsigned chr = *start; \
+ if (!ws) { \
+ if (!DO_IT_SPACECHECK(chr)) \
+ break; \
+ } else { \
+ char *p = ws; \
+ do { \
+ if (*p == chr) \
+ goto lead##TYPE; \
+ } while(*++p); \
+ break; \
+ } \
+lead##TYPE:; \
+ } \
+ for (; start < end; start++) { \
+ unsigned chr = *start; \
+ if (!ws) { \
+ if (DO_IT_SPACECHECK(chr)) \
+ if (foundspace) \
+ continue; \
+ else \
+ foundspace=1,chr=' '; \
+ else \
+ foundspace=0; \
+ } else { \
+ char *p = ws; \
+ do { \
+ if (*p == chr) \
+ if (foundspace) \
+ goto skip##TYPE; \
+ else { \
+ foundspace=1;chr=*ws; \
+ goto copy##TYPE; \
+ } \
+ } while(*++p); \
+ if (foundspace && (chr=='\n' || chr=='\r')) { \
+ dst[-1] = chr; foundspace=0; \
+ goto lead##TYPE; \
+ } \
+ foundspace=0; \
+ } \
+copy##TYPE: \
+ *dst++ = chr; \
+skip##TYPE:; \
+ } \
+ len = dst - (TYPE*)sb.s->str; \
+ }
+ case 0: DO_IT (p_wchar0); break;
+ case 1: DO_IT (p_wchar1); break;
+ case 2: DO_IT (p_wchar2); break;
+#undef DO_IT
+#undef DO_IT_SPACECHECK
+ }
+ if (foundspace)
+ len--;
+ sb.s->len = len;
+ RETURN finish_string_builder (&sb);
+}
+
/*! @decl string trim_all_whites (string s)
*! @belongs String
*!
diff --git a/src/builtin_functions.h b/src/builtin_functions.h
index 36ef712705638d1e775a4db2c8f1cf19f23f747d..90e959732924659bbd1b5326a8ccd0d4cf8d579e 100644
--- a/src/builtin_functions.h
+++ b/src/builtin_functions.h
@@ -2,7 +2,7 @@
|| This file is part of Pike. For copyright information see COPYRIGHT.
|| Pike is distributed under GPL, LGPL and MPL. See the file COPYING
|| for more information.
-|| $Id: builtin_functions.h,v 1.40 2009/11/12 15:10:21 grubba Exp $
+|| $Id: builtin_functions.h,v 1.41 2010/01/02 13:17:10 srb Exp $
*/
#ifndef BUILTIN_EFUNS_H
@@ -181,6 +181,7 @@ PMOD_EXPORT void f_ctime(INT32 args);
PMOD_EXPORT void f_mkmapping(INT32 args);
PMOD_EXPORT void f_string_count(INT32 args);
PMOD_EXPORT void f_string_trim_whites(INT32 args);
+PMOD_EXPORT void f_string_normalize_space(INT32 args);
PMOD_EXPORT void f_string_trim_all_whites(INT32 args);
PMOD_EXPORT void f_program_implements(INT32 args);
PMOD_EXPORT void f_program_inherits(INT32 args);